Temporary Loss Of Network On Database Network Interface When Oracle Database Starts Up (Doc ID 2455332.1)

Last updated on JUNE 02, 2020

Applies to:

Symptoms

When starting an Oracle Database on a system running Oracle Linux, the network is lost for roughly 10 seconds and is then restored.

A tcpdump looks similar to the following:

• ping request and reply initially works as expected:
  
  21 2018-09-19 14:35:00,494028 10.xx.xxx.133 10.xx.xxx.201 ICMP 102 Echo (ping) request id=0x890e, seq=98/25088, ttl=64 (reply in 22)
  22 2018-09-19 14:35:00,494063 10.xx.xxx.201 10.xx.xxx.133 ICMP 102 Echo (ping) reply id=0x890e, seq=98/25088, ttl=64 (request in 21)
  
  
• then there are connection resets observed:
  
  23 2018-09-19 14:35:00,712488 10.xx.xxx.201 10.xx.xxx.133 TLSv1.2 275 Application Data
  24 2018-09-19 14:35:00,712627 10.xx.xxx.133 10.xx.xxx.201 TCP 70 29069 → 3872 [ACK] Seq=1 Ack=206 Win=1027 Len=0 TSval=442483460 TSecr=790283494
  26 2018-09-19 14:35:00,714356 10.xx.xxx.201 10.xx.xxx.133 TCP 58 7 → 48296 [RST, ACK] Seq=1 Ack=1 Win=0 Len=0
  
  
• Followed by roughly 8 seconds silence and then following is logged:
  
  30 2018-09-19 14:35:12,494094 IntelCor_e2:9c:30 IntelCor_d9:d6:f2 ARP 46 10.xx.xxx.201 is at 90:e2:ba:e3:9a:30
  
  
• after which, ping resumes:
  
  31 2018-09-19 14:35:12,494187 10.xx.xxx.133 10.xx.xxx.201 ICMP 102 Echo (ping) request id=0x890e, seq=110/28160, ttl=64 (reply in 32)
  32 2018-09-19 14:35:12,494222 10.xx.xxx.201 10.xx.xxx.133 ICMP 102 Echo (ping) reply id=0x890e, seq=110/28160, ttl=64 (request in 31)

Cause

My Oracle Support provides customers with access to over a million knowledge articles and a vibrant support community of peers and Oracle experts.